Python. Python. 1 Python. M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

Transcription

1 Python 1 Python M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

2 Python makes you fly M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

3 Let s start ipython vs python ipython Note that you have an access to the OS shell ls pwd reverse search: ctrl + R Automatic completion works! use Tab Comments with # Great help use? or help help pow pow? M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

4 ipython II History arrow up (shows previous command in history) arrow down (shows next command in history) po arrow up (shows previous command starting with po) history # (prints the complete recorded history) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

5 Array basics You need numpy import numpy as np Alternatively, you can use (less safe) from numpy import * Several possibilities how to create an array x=np.zeros((20,30)) Creates a 20x30 array of zeros x=np.ones((20,30)) x=np.arange(10) Creates integer arrays with sequential values Starting with 0 x=np.arange(10.) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

6 Array basics II x=np.arange(3,10) x=np.arange(3.,10.,1.5) x=np.array([3.,1.,7.]) Creating arrays form literal arguments x=np.array([[3.,1.][1.,7.]]) Be careful about reals and integers! x=3 y=4 x/y x=3. y=4 x/y Array indexing starts from 0! x,x[0],x[1] M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

7 Array types Array numeric types The default for integer is usually 32-bit integers (in numpy called int32) The default for floats is 64-doubles (in numpy called float64) How to find out the type of an array x with dtype x=arange(4.) x.dtype Converting an array to a different type with adtype() y=x.astype(float32) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

8 Array operations x.max() x.min() x.std() x.mean() x1=x.copy() x-=x.mean() equivalent with x=x-x.mean() max(x) min(x) std(x) mean(x) x1=copy(x) x-=mean(x) equivalent with x=x-mean(x) First and last 10 entries of an array x = np.random.random(100) x[:10] x[-100:] x[2:5] M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

9 Multidimensional arrays Creating an N M array N = 5 M = 3 x2 = np.zeros((n,m)) Size and shape of an array size(x2) shape(x2) Reshaping of an array x3 = np.arange(10).reshape(2,5) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

10 Linear algebra import numpy.linalg as la An inverse of a matrix x = np.array([[1,1],[1,-1]]) la.inv(x) dot(x,la.inv(x)) Finding an eigenvalues and vectors x = np.array([[1,2],[2,1]]) val, vec = la.eig(x) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

11 Pyhton For statements for i in range(3): x=i*i print x Indentation determines a block of code (such as for) cycle In interactive mode starting a block causes the interpreter to prompt with... (typing an empty line with enter terminated the block) Tabs may be used for indentation, but their use is not recommended M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

12 Python If statements x=0 if x==0: print "x equals 0" elif x==1: print "x equals 1" else: print "x equals something else than 0 or 1" M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

13 Plotting - handled with matplotlib module import matplotlib.pyplot as plt import numpy as np Plotting random series of y values y = np.random.random(50) plt.figure() plt.plot(x) plt.show() M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

14 Plotting II Plotting x against y values t = np.arange(10) y1 = np.random.random(10) y2 = np.random.random(10) plt.figure() plt.plot(t,y1, r ) plt.plot(t,y2, go- ) plt.show() Clear figure plt.clf() Close all figures plt.close( all ) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

15 Other modules useful in Earth sciences For seismologist obspy module M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

16 Geochemical systems - one reservoir The concentration C i of an element i must follow (in the simplest case) dc i dt = C i τ Concentration in the box is homogeneous Initial condition C i (t = 0) = C 0 M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

17 Ordinary differential equations (ODE) We will need SciPy (Scientific tools for Python) import scipy as sp from scipy.integrate import odeint The function odeint finds solution of equation dy dt = f(y, t) With initial conditions y(0) = y 0 Where y is a length N vector and f is a mapping from R N to R N M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

18 Make your first Python script touch box model single.py #!/bin/python import numpy as np from scipy.integrate import odeint import matplotlib as mpl import matplotlib.pylab as plt Parameters of the system tau=1 y0=1 # initial conditions M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

19 ODE II Define times when you want to find a solution t = np.arange(0,10., 1.) Right hand side of the equation def func(y,t): return -y/tau Solve the system yy=odeint(func,y0,t) # solution of dc/dt=... M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

20 ODE III We know analytical solution time = np.arange(0,10.,0.1) concentration=y0*np.exp(-time/tau) # analytical solution Plotting plt.figure() plt.title( Concentration time evolution ) plt.ylabel( Concentration ) plt.xlabel( Time ) plt.plot(t,yy, ro ) plt.plot(time,concentration) Run your script in ipython execfile( box model single.py ) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

21 Box model - n reservoirs with concentrations C i Coltice at al. (GRL, 2000) Transport of an inert gas in the mantle Reservoir 6 Reservoir 5 Degassing K_n=K_top The n equations to solve dc i dt = v zn L (K i 1C i 1 K i C i ) K i = 1 i 1 or n Reservoir 4 Reservoir 3 Reservoir 2 Reservoir 1 K_0 C_0=K_bot C_n Reinjection K 0 C 0 = K bot C n = 0.2C n K n = K top = 10 v z = 1 mm/y (vertical velocity) n number of boxes L size of the domain M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

22 Box model with Python touch box model.py #!/bin/ipython import numpy as np from scipy.integrate import odeint import matplotlib as mpl import matplotlib.pylab as plt Parameters of the system vz=0.001 # m/year ktop=10.0 kbot=0.2 nn=100 # number of boxes LL=3E6 # size of mantle [m] M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

23 Box model with Python II t = np.arange(0,1.8e9, 1E6) # times - min, max, step y0=np.ones(nn) # array of initial concentration ynew=np.zeros(nn) def func(y,time): for ii in range(nn): if ii==0: # bottom ynew[0]=kbot*y[nn-1]-y[0] elif ii==nn-1: # top ynew[nn-1]=y[nn-2]-ktop*y[nn-1] else: ynew[ii]=y[ii-1]-y[ii] alpha=vz*ll/nn return alpha*ynew yy=odeint(func,y0,t) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

24 Plotting the results zz=np.arange(0.,1.,1./nn) zz=np.append(zz,1.) ntime=size(t)-1 print plotting at time:, t[ntime] plt.figure() plt.title( Concentration profile ) plt.xlabel( Concentration ) plt.ylabel( Height ) plt.plot(np.append(yy[ntime],yy[ntime][nn-1]),zz,drawstyle= steps ) plt.axis([0,max(yy[ntime]),0,1]) plt.savefig( fig1.pdf ) M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25

25 Plotting the results II: Animation files=[] fig=plt.figure(figsize=(5,5)) ax=fig.add subplot(111) jj=0 for ii in range(0,size(t),size(t)/20): # 20 frames jj=jj+1 ax.cla() ax.plot(np.append(yy[ii],yy[ii][nn-1]),zz,drawstyle= steps ) fname = fig%03d.png %jj print Saving frame, fname fig.savefig(fname) files.append(fname) os.system( mencoder mf: tmp*.png -mf type=png:fps=10 / -ovc lavc -lavcopts vcodec=wmv2 -oac copy -o animation.mpg ) mplayer M.Ulvrova, L.Pouilloux (ENS LYON) Informatique L3 Automne / 25